Thursday, October 01, 2009

Spectrum Databases (Not)

The idea of using a database to regulate radio operation – or, to “control access to spectrum”, to use the S-word nomenclature – has been gaining ground.

For example, both Michael Calabrese (The End of Spectrum ‘Scarcity’, New America Foundation Wireless Future Program Working Paper No. 25, June 2009) and Kevin Werbach (Castle in the Air: A Domain Name System for Spectrum, TPRC September 2009) have argued that the database(s) contemplated to manage device operation in the TV white spaces could be the foundation for a method to increase the amount of radio operation.

Thinking through how such a database might be used shows the advantage of approaching radio regulation as coordinating operations, rather than using conventional approach of “dividing up spectrum”.

The regulatory challenge is therefore not "spectrum databases" but "radio operation databases".

In a first approximation – and perhaps even as the ultimate solution, if one uses the “spectrum” approach – a database would be a listing of “vacant” channels; a device would query the database for “available” channels, and operate in one. When one starts from the basis that spectrum is an asset like land to be divided up and distributed, vacancy is a self-evident concept; it derives from the attributes of the underlying asset, and not by reference to the intended use.

However, context is everything in radio operation. Whether harmful interference will result from the operation of an added radio system depends not only on its transmissions, but also the transmit and receive characteristics of the incumbent system.

Consider, for example, three channels: A, B, and C. Let’s say incumbent system #1 operates using channel A. Channels B and C are nominally vacant. Can an incoming system #2 operate in those channels? If both system #1 and #2 use traditional cellular technology (i.e. FDM, e.g. 3G), the answer is yes. But if #1 uses 3G and #2 uses TDM technology like WiMAX, then the answer is No: there needs to be a guard band between them, and system #2 can only use Channel C. Channel B needs to be left “vacant”. (This is a live issue: see e.g. Ars Technica on the argument between T-Mobile and M2Z over the rules for the AWS-3 band band.)

A mental model informed by spectrum-as-land is therefore not an ideal guide to understanding what needs to be in the database. (More generally, one needs to refine the metaphor to better guide regulation, as Weiser and Hatfield did last year by introducing the concept of "zoning the spectrum" in Spectrum Policy Reform and the Next Frontier of Property Rights, 15 Geo. Mason L. Rev. 549.)

An approach grounded in coordinating operations, on the other hand, leads to the understanding that what needs to be in the database is not just a frequency range and geographic region, but all the relevant parameters of an incumbent operation. The short list would add receiver performance (ability to reject interference) and duty cycle (near-constant transmission like cellular systems, vs. very intermittent but intense uses like firefighting) to the usual suspects of transmitter location, emitted power, and transmit mask.

The task is not to find a “vacant channel”, but to determine if an incoming operator will cause harmful interference. This requires, in addition to the operating parameters of the incumbent and incoming systems, information about the spatial distribution of incumbent and incoming radios, and a propagation model to connect the two.

Ofcom is the regulator that has thought most deeply about ways to better characterize the interference characteristics of radio systems; see e.g. Ofcom’s Guide to Spectrum Usage Rights (SURs) and William Webb’s recent paper Licensing Spectrum: A discussion of the different approaches to setting spectrum licensing terms.

A well-founded framework for generalizing the white space database – where interference management between incumbents and new entrants hard-coded into the FCC rules for white space device operation – could benefit from new radio operating metaphors (grind axe: see my De-situating spectrum: Rethinking radio policy using non-spatial metaphors, DySPAN 2008) and the application of a SUR-like approach.

One will also have to think carefully about the minimal set of parameters needed to facilitate interference avoidance, since it's easy but economically inefficient to come up with a very long list of attributes that describe radio operations. The Silicon Flatirons Center recently examined this issue in a summit on defining out-of-band operating rules.

No comments: